While significant progress has been made toward identifying effective interventions for preschool-age children with autism (National Research Council, 2001), few scientifically rigorous studies have compared active ingredients of these interventions or examined outcomes focused on core deficits. To address these areas of need, this collaborative, multi-site project combines the expertise of investigators experienced in randomized controlled clinical trials (RCTs), in the study of core deficits in young children with autism, and in data management and analysis of multi-site clinical trials. The goal of this project is to compare the efficacy of two interventions for improving spoken language and reducing symptoms of autism: (1) Discrete trial training (DTT)--an applied behavior analysis approach emphasizing highly structured teaching of school readiness skills (match-to-sample, imitation, functional play, and receptive and expressive language) and (2) Interpersonal developmental approach (IDA): a visually supported, child-focused, flexible engagement, socialcommunicative engagement approach on joint attention, symbolic play, and the use of conventional symbols within socially valid communicative contexts. Each site will recruit 64 children with autism (3-4 years of age; total sample=192) who are engaged in 25 hour per week early intervention programs. Children will be randomly assigned to DTT or IDA. In each condition, children's ongoing early intervention programs will be augmented with two 30-minute sessions daily of the study intervention (DTT or IDA) conducted by supervised therapists for 4 months, with transition to home therapy for 2 months. The primary aims are 1) to compare intervention conditions (DTT and IDA) on child outcomes of spoken language at the end of treatment and 6 months later, 2) to compare intervention conditions on child outcomes of core deficits of social features of autism at end of treatment and 6 months later, and 3) to examine potential moderators (e.g., mental age, language age) on treatment outcome. Secondary exploratory analyses focus on potential mediators (parent synchronization of joint attention and changes in parental expectancies), and non-specific factors (quality and quantity of community treatment). This study addresses multiple priorities in the autism research matrix (National Institute of Mental Health, 2004), including multisite RCTs of early intervention, isolation of active ingredients, identification of mediators or moderators, and treatment of core deficits.